2. Materials
2.2 Devices
Device Manufacturer
Microtome Leica SM2000R (Ref. No.
1464/11.1998, Ser. No.1469) Leica Mikrosysteme GmbH, Wetzlar
pfm Waterbath (Ref. Nr. 041010, Ser. No.
300359) Pfm medical AG, Köln
pfm cooling plate (Art. Nr. 501003, Ser. No.
15020001) Pfm medical AG, Köln
Tissue-Tek Prisma automated slide stainer Sakura Finetek Europe B.V., AJ Alphen aan den Rijn
Autostainerlink48 Dako, An Agilent Technologies
Company, Hamburg IHC advanced staining system Dako Omnis, Hamburg
Zeiss microscope (SIP 51444) Axiophot Carl Zeiss Microscopy GmbH, Jena
Olympus BX46F microscope Olympus, Shinjuku, Tokyo Olympus BX53 microscope Olympus, Shinjuku, Tokyo peqSTAR Thermocycler, Peqlab VWR, Avantor, Radnor,
Pennsylvania
Thermomixer Comfort Eppendorf AG, Hamburg
Qubit fluorometric quantitation Thermo Fischer Scientific, Waltham, Massachusetts
InnuPure C16 Touch Analytik Jena AG, Jena
QIAxcel Advanced System, Capillary
Electrophoresis Qiagen, Hilden
MiSeq System Illumina, San Diego, California
Centrifuge 5430R Eppendorf AG, Hamburg
Vortexer Genie 2 Bender & Hobein GmbH, Zürich 2.3 Chemicals and Reagents
Chemical/Reagent Ref./Art. No. Manufacturer
Clearify Clearing Agent NC9837230 American MasterTech, Lodi, California
EnVision Flex Target Retrieval Solution, pH low (50x)
FV805
DM849 Dako, Hamburg, Germany EnVision Flex Target
Retrival Solution, pH high (50x)
DM848 Dako, Hamburg, Germany
Wash Buffer GC807
DM851 Dako, Hamburg, Germany
Chemical/Reagent Ref./Art. No. Manufacturer EnVision Flex
Peroxidase-Blocking Reagent DM841 Agilent, Santa Clara, California EnVision Flex Substrate
Buffer DM843 Agilent, Santa Clara, California
EnVision Flex+ Mouse
Linker SM804 GV821
DM844 Agilent, Santa Clara, California EnVision Flex+ Rabbit
Linker SM805 GV809
DM845 Agilent, Santa Clara, California EnVision Flex/HRP DM842 Agilent, Santa Clara, California EnVision Flex Substrate
Working Solution DAB+
Chromogen
DM847 Agilent, Santa Clara, California
Shandon Eosin Y 6766010 Thermo Scientific, Waltham, Massachusettes
Hematoxylin 7211 7211L Thermo Scientific, Waltham, Massachusettes
Ethanol 99% 2294.5000 Chemsolute, Th. Geyer GmbH &
Co. KG, Renningen, Germany
Xylol 371.5000 Chemsolute, Th. Geyer GmbH &
Co. KG, Renningen, Germany Ethanol 96% 2293.5000 Chemsolute, Th. Geyer GmbH &
Co. KG, Renningen, Germany DNAse/RNAse free water 2352744 B. Braun, Melsungen
PhiX Control v3 FC-110-3001 Illumina
EB Puffer 19086 Qiagen
Milli-Q Direct 8 Water
Purification System ZR0Q008WW Merck Millipore, Darmstadt 2.4 Further Materials
Microscope slides Thermo Scientific, Braunschweig
Süssefrost microscope slides Süsse Labortechnik GmbH, Gudensberg Voyager Adjustable Pipette Integra Biosciences AG, Zizers, Schweiz Tacta Pipettes (10µl, 100µl, 1000µl) Sartorius, Göttingen
PCR Plates 4ti-0750/TA/IND 4titude Wotton, Surrey Magnet Plate
2.5 Computer Software and Internet Databases
Description Application
Circos 0.69 Software Visualization of genomic data CLC Genomics workbench 11, Qiagen,
Hilden Evaluation of NGS data
ClinVar archives
https://www.ncbi.nlm.nih.gov/clinvar/ Mutation database COSMIC Catalogue of Somatic Mutations
http://cancer.sanger.ac.uk/cosmic Mutation database MEDOS Web 9.3 Rev. 1424 Pathology records
Microsoft Excel 2016 Configuration of NET database, Statistics, construction of diagrams
Microsoft Word 2016 Thesis writing
http://www.ncbi.nlm.nih.gov Pubmed
Statistica 13.3 Statistics, construction of graphs ZEN 2012 (blue edition), version 1.1.2.0,
Carl Zeiss microscopy GmbH 2011 Microscope photographs
2.6 Kits
Kit Manufacturer
PTLink DM 828 Dako, An Agilent Technologies
Company, Hamburg
PTLink DM 829 Dako, An Agilent Technologies
Company, Hamburg
QIAseq Index Kit QIAGEN, Hilden
GeneRead DNAseq Panel PCR Kit V2
(Human Myeloid Neoplasms Panel) QIAGEN, Hilden GeneRead DNAseq Custom Panel V2
(181902 CNGHS-00156X-205) QIAGEN, Hilden GeneRead DNAseq Targeted Panel V2
(3027978 NGHA-102X-96) QIAGEN, Hilden
GeneRead DNA Library I Core Kit 180434 QIAGEN, Hilden GeneRead DNA Amp Kit 180455 QIAGEN, Hilden
MiSeq Reagent Kit v2 MS-102-2002 Illumina, San Diego, California innuPREP FFPE DNA Kit- IPC16 Analytik Jena AG, Jena
Qubit Assay Q32854 Thermo Fischer Scientific
AmPure Beads A63881 Beckmann Coulter, Brea, California BIOO BC Adapter and Primer Mix NOVA-514102
2.7 Antibodies Antibody Ref.
Number Species Manufacturer Concentration Anti-DAXX HPA008736 rabbit Sigma-Aldrich, St. Louis,
USA 0.1 mg/ml,
1:200 Anti-ATRX HPA001906 rabbit Sigma-Aldrich, St. Louis,
USA 0.2 mg/ml,
1:500 Anti-SDHA HPA041981 rabbit Sigma-Aldrich, St. Louis,
USA 0.3 mg/ml,
1:200 Anti-SDHB HPA002868 rabbit Sigma-Aldrich, St. Louis,
USA 0.2 mg/ml,
1:200 Anti-EZH2 6034735 mouse Novocastra Leica
Biosystems, Newcastle, United Kingdom
20 mg/L, 1:50
Anti-H3K27me3 9733S
P68431 rabbit Cell Signaling Technology,
Massachusetts, USA
1:500
Anti-gH2AX P16104
9781T rabbit Cell Signaling Technology,
Massachusetts, USA
1:100
Anti-c-myc Ab32072 rabbit Abcam, Cambridge 0.203 mg/ml, 1:100
Anti-Yap1 P46937
14074S rabbit Cell Signaling Technology,
Massachusetts, USA
1:400
Anti-Pax 8 363M-18 mouse Cell Marque, Rocklin, California
Anti-TTF1 6041484 mouse Novocastra Leica
Biosystems, Newcastle, United Kingdom
1:100
Anti-CDX2 GA080 mouse Dako Flex, Carpinteria,
CA RTU
Anti-Serotonin M0758 mouse Dako Agilent Pathology
Solutions, Hamburg 90 mg/L, 1:100 Anti-SSTRA2 RBK046-05 rabbit Zytomed Systems,
Berlin 1:100
Anti-CD5 IR082 mouse Dako Flex, Carpinteria,
CA, USA RTU
Anti-CD117 A4502 rabbit Dako Agilent Pathology
Solutions, Hamburg 12.2 g/L, 1:500 Anti-Calcitonin A0576 rabbit Dako Agilent Pathology
Solutions, Hamburg RTU
Antibody Ref.
Number Species Manufacturer Concentration Anti-Keratin Z0622 rabbit Dako Agilent Pathology
Solutions, Hamburg RTU
Anti-Chromogranin A
238M-90 mouse Medac GmbH, Wedel RTU
Anti-Ki-67
Clone MIB-1 IR626 mouse Dako Denmark A/S,
Glostrup RTU
Anti-RB1 HPA050082 rabbit Sigma-Aldrich, St. Louis,
USA 0.1 mg/ml;
1:100 Anti-P53 Clone DO-7 mouse Dako Agilent Pathology
Solutions, Hamburg RTU
3. Methods
3.1 Patient Characteristics
Clinical information regarding age, sex, relevant pre-existing conditions, date of initial NET diagnosis, age at diagnosis and tumor size was obtained from the pathology medical records system Medos. Furthermore, resident registration offices were contacted to obtain information on patients’ decease dates, if applicable, for survival statistics.
3.2 Tumor Tissue Preservation and Preparation
All tissues were formalin fixed and paraffin embedded (FFPE) archival materials. For HE and unstained paraffin slides, the FFPE blocks were cut with a microtome into two micrometer sections and were mounted on microscope slides for further analysis.
3.3 Tissue-Microarray (TMA)
Tissue-microarray blocks were manufactured by Targos Molecular Pathology GmbH in Kassel, Germany. First, HE slides of each tumor were viewed under a microscope to determine and mark two tumor hotspots. HE microscope slides and the associated tumor blocks were sent to Targos GmbH. Here, tissue cores were punched out of the donor tumor blocks at the corresponding spots on the marked HE slides. The two-millimeter tissue cores were placed in receptor blocks. Tissue extraction and induction were performed with the TMA Master 3D HISTECH. 60 tumor samples were fit onto one tissue-microarray block. There were two samples for each tumor. The finished TMA blocks were sent back to the University Medical Center Göttingen, where these were mounted on microscope slides.
3.4 HE Stains
HE Stains were performed by the HE-staining machine Tissue Tek Prisma, on microscope slides mounted with the tumor tissue. Stains were conducted by the immunohistochemistry lab of the University Medical Center Göttingen. The staining procedure is as follows:
The duration of each incubation lasted two minutes per cycle. If not further specified, each incubation represents one cycle. First, the tumor slides were incubated in Xylol for one cycle, followed by abs. alcohol for two cycles, alcohol 96% for once cycle, alcohol 75%, distilled water, and hemalum for the three cycles. The tumor slides were then rinsed with warm running water for two minutes and again incubated in eosin 2%
for two cycles, distilled water for one cycle and alcohol 96% for two cycles. The slides are then rinsed with abs. alcohol for two minutes and incubated in xylol for three cycles.
Lastly, slides are covered with a film coverslipper.
3.5 Immunhistochemical Stains
Immunohistochemical stains were performed on tissue micro array sections on the Dako Omnis advanced staining system and on the Dako Autostainer Link 48. The staining protocol is as follows:
First, the tissue specimens were deparaffinized in clearify clearing agent for one cycle of one minute and rehydrated with DI water for one cycle of five seconds. Then epitopes were retrieved with EnVision Flex Target Retrieval Solution, pH low (TTF1, ki67, DAXX, ATRX, SDHA, Calcitonin, RB1) or high (CDX2, CD5, CD117, chromogranin A, SDHB, EZH2, H3K27me3, γH2AX, c-Myc, YAP1, Serotonin, SSTR2A, Keratin, p53), depending on the IHC stain, for 30 minutes. After washing the sections with a wash buffer for two cycles of 2:40 minutes, sections were incubated in the primary antibody for 20 to 30 minutes. Again, the sections were washed with a wash buffer for 10 cycles each consisting of two minutes, then incubated for three minutes in EnVision Flex Peroxidase-Blocking Reagent. Following this, sections were washed with the wash buffer for 10 cycles of two minutes and incubated in EnVision Flex/HRP, marked secondary polymeric antibodies. After washing the slides for two minutes for 20 cycles, washing with distilled water for 31 seconds, and again washing with wash buffer for two minutes and 10 cycles, the slides were finally incubated in EnVision Flex Substrate Working Solution, a substrate chromogen which allows visualization of the immunoprecipitants. Lastly, after another cycle of washing with wash buffer and distilled water, the slides were counterstained with hematoxylin for three minutes and analyzed under a light microscope.
3.6 IHC Evaluation
The immunohistochemistry stains were evaluated using a two-tiered and a three-tiered score. Stain intensity and the percentage of positive stained tumor cells were assessed under 10x and 40x microscope objectives. Intensity scores ranged from 0 (no staining) to 1 (weak staining) to 2 (strong staining). The percentage of positive staining tumors cells were scored as <25%, <50%, and >50%. This resulted in the following evaluation possibilities: 0, 1<25, 1<50, 1>50, 2<25, 2<50, 2>50. Each tumor received a score for each immunohistochemical stain. For statistical analysis we
grouped the scores into negative expression (0, 1<25, 2<25) and positive expression (1<50, 1>50, 2<50, 2>50) or negative expression (0, 1<25), weak expression (2<25, 1<50), and positive expression (1>50, 2<50, 2>50).
3.7 Ki67 Labeling Index Evaluation
The proportion of ki67 positive cells was determined by viewing tumor sections under a light microscope using a 40x objective (field-of-view diameter of 0.55 mm, resulting in 10 HPF = 2.37 mm2). Cells were analyzed by eyeballing and estimating the percentage of positive cells. Second, a digital image analysis counter (morphometric analysis) was used, which evaluated the percentage of positive cells with a standard deviation range. Cases in which there was an obvious discrepancy between the morphometric data and an estimate by eyeballing, a photograph of a representative hotspot area was taken at 400fold magnification, printed out, and positive versus negative tumor cell nuclei were manually counted.
3.8 WHO, ENETS, and Pelosi et al. Classification
The HE slides of 72 thymic NET were reviewed for morphologic features and diagnosis under a light microscope. Thymic NET were graded according to the WHO 2015 grading scheme. The cutoffs were: <2 mitoses per 2 mm² and no necrosis for TC; 2-10/2 mm², with or without necrosis for AC; and >2-10/2 mm², with or without necrosis for large cell NEC and small cell carcinoma. Mitotic counts were assessed in 10 HPF on HE-stained sections, using an Olympus BX53 microscope (40x objective, field-of-view diameter of 0.55 mm, resulting in 10 HPF = 2.37 mm2). Only unequivocal mitoses were counted.
56 of these cases were also classified according to the ENETS WHO classification of gastro-entero-pancreatic neuroendocrine tumors. The same cutoff rates that apply to GEP-NET, were used to grade the TNET: TC ≤ 3%, AC 4-20%, and HGNEC > 20%.
60 TNET were classified according to the PNET classification proposed by Pelosi et al. This classification is based on the ki67 LI, mitotic count and necrosis. All three parameters are tiered based on three different expression levels, which groups the tumors into G1, G2 and G3 as shown in Table 2 in chapter 4.3.1.
3.9 DNA Isolation
Tumor tissue was extracted from the FFPE tumor blocks and centrifuged at maximum speed for one minute. Lysis Solution BC and Proteinase K were added to the sample and mixed vigorously. The mixture was then incubated at 65 °C for one hour in a
thermal mixer. After the lysis step, solution QPS was added to the sample and again centrifuged for one minute. The mixture was incubated in another thermomixer at 90
°C for one hour. MAG Suspension F and the lysed DNA sample were transferred into the reagent strips of the InnuPure C16. Once the protocol was completed by the InnuPure C16 system, the DNA was extracted in Elution tubes. This process was carried out by the Molecular Pathology Lab of the University Medical Center Göttingen.
3.10 Next-Generation Sequencing
NGS was performed as a courtesy of S. Küffer (Institute of Pathology, University Medical Center Göttingen). DNA was isolated from eleven LCNEC following the protocol listed in chapter 3.10. The DNA was quantified using the Qubit Assay (Thermo Fisher) and the concentration of every sample was adjusted to 2,5 ng/ml. Multiplex PCR was performed using the Human Myeloid Neoplasms Sequencing Panel (Cat. No. NGHS-003) of the GeneRead DNAseq Panel PCR Reagent V2 (Qiagen) according to the manufacturers protocol. DNA amplicons were purified with the AmPure Beads (Qiagen). DNA was end-repaired and adenylated (A-addition) using the GeneRead DNA Library I Core Kit (Qiagen). The adaptors (Illumina) were ligated using the GeneRead DNA Library Core Kit (Qiagen) followed by a purification step with AmPure Beads. The libraries were then size-selected, and PCR-amplified to increase sequencing depth using the GeneRead DNA Amp Kit (Qiagen). After an additional purification with the AmPure Beads, library sizes and concentrations were measured with QIAxcel capillary electrophoresis (Qiagen). The library was then diluted, pooled and denatured for subsequent sequencing on the Mi-Seq system using the MiMi-Seq Reagent Kit v2 (Illumina MS-102-2002).
For data analysis, the FastQ files were analyzed in the CLC Biomedical Workbench (Qiagen) using an in-house workflow. The reads were mapped on hg19 (human genome 19, National Center for Biotechnology Information build 37) followed by an initial variant calling. Subsequently, local realignments, primer clipping, and low-frequency variant calling were performed. False positives were removed based on the read quality and the forward/reverse balance. All variants called were checked manually for sequencing artefacts. The average coverage was >500 in all samples, the mutations had at least 50 variant reads.
3.11 Shallow Whole Genome Sequencing by Chronix GmbH
This analysis was performed by Chronix GmbH (Dr. Julia Beck and Prof. Ekkehard Schütz, Göttingen, Germany). Molecular analyses were possible in 63 tumor samples (13 TC, 30 AC, 16 LCNEC, 4 SCC). Extracted DNA was ultrasonically sheared to an approximate fragment size of 200 bp using a Covaris S2 focused-ultrasonicator.
Sequencing libraries were prepared using the NEBNext Ultra II DNA Library Preparation Kit for Illumina (New England Biolabs, Frankfurt, Germany) according to the manufacturer’s instructions. Paired-end sequencing (37/38 bp) was conducted on an Illumina NextSeq500 with base calling using the bcl2fastq program version 2.17.1.14. An average of 20.2 M (STDEV: 5.5M) reads were generated per sample.
Sequences were mapped to the human reference genome (HG19) using the BWA version 0.7.12 (average of mapped reads: 15.7 M, STDEV: 6.5 M) (Li and Durbin 2009).
Copy-number analysis based on read-count data was conducted using the QDNAseq R package (version 1.10.0) using a fixed window size of 500 kbp (4407 windows in total) (Scheinin et al. 2014). The obtained log2 ratios were smoothed by applying the circular binary segmentation algorithm using the R package Copynumber version 1.14.0 (Nilsen et al. 2012). Based on the absolute log2 ratios observed in the four normal samples (mean + 10*STDEV) the thresholds for calling copy-number gains/losses in the tumor samples were set to of 0.09/-0.09, respectively.
3.12 CNI Test by Chronix GmbH
Library Preparation and Sequencing: Illumina paired-end libraries were generated using DNA extracted from the TNET mentioned above. In brief, fragmented DNA was end-repaired, A-tailed and multiplex sequencing adapters were ligated (NEBNext Ultra II, New England Biolabs). After amplification of adapter-ligated fragments (9-19 cycles) the samples were sequenced on an Illumina NextSeq500 according to manufacturer’s protocols.
After de-multiplexing, the sequence data were mapped to HG19. Duplicate reads were removed using the Picard Tools and low-quality reads (mapping quality <60) were removed using Samtools.
Depth of coverage analysis was performed, in which log2 normalized read count ratios were calculated in sliding windows for each tumor, for local reoccurrence and for metastasis. CNV-Seq was used with windows sizes set to either 500 kbp or 5 Mbp.
The significance limits were set to a read count ratio of 0.09 and -0.09. These values were used to color significant copy number imbalances in the Circos plots using the Circos software.
A chromosomal instability (CNI) score was used for the comparison between samples and between groups of samples (Schutz et al. 2015). In brief, after sequencing, the mapped reads are counted in windows along the chromosomes. A z-score was then calculated for each window. The CNI score is obtained from all significantly aberrant windows (windows above/below the thresholds) by statistical comparison to a normal population.
Genes, for which causative impacting deletions or amplifications have been described in the COSMIC Cancer Gene Census database, were extracted for regions with gains/losses present in >10%, >20% and >25% in the TC, AC and LCNEC/SCC groups, respectively (Futreal et al. 2004). The percentages of windows above/below the thresholds (copy-number instability, CNI score) were calculated as a general measure of the amount of copy-number aberration present in each tumor.
3.13 Statistical Analysis
Statistica version 13.3 (Dell, Germany) was used for statistical analyses. Survival analyses were performed via the Kaplan-Meier method. Statistical significance was tested with the log-rank test. Correlations between metric data were configured with scatterplots and the strength of correlation was assessed with correlation coefficient.
To demonstrate the association of grading system subgroups with genetics, boxplots were created. Statistical significance was tested with the Kruskal-Wallis ANOVA test.
Associations between potential survival predictors and survival were made with Cox hazard regression ratios. The correlation between immunohistochemistry and tumor subgroups was tested with Mann-Whitney U Test, and Spearman Rank Test. P-values
<0.5 were regarded as significant.
4. Results
4.1 TNET Cohort and Clinical Features
Thymic neuroendocrine tumor cases were gathered from the University Medical Center Göttingen Pathology archives and from international cooperating institutions mentioned in chapter 2. Information regarding patient characteristics was obtained from the Medos database. Survival times were acquired from resident registration offices. The clinical features of the TNET cohort used in this study are summarized in Table 1. Patient characteristics.
Table 1: Patient characteristics.
With a total of 107 cases, 22 (20.5%) were classified as TC, 51 (47,6%) as AC, 28 (26%) as LCNEC and 6 (5.6%) as SCC according to the WHO 2015 classification.
There was a male predominance in this cohort: 78 patients were male, 27 patients were female and 2 were unknown. The median age of all patients was 54 at time of diagnosis. Among the 107 cases, 51 tumors (7 TC, 32 AC, 10 LCNEC and 3 SCC) had available survival data. In all, at the time of data collection 25 patients were deceased and 26 were alive. Of 7 TC patients, 6 (85%) were alive and 1 (17%) had died. Of 31 AC patients, 18 (61%) were alive and 13 (16%) had died. Of 10 LCNEC patients, 2 (20%) were alive and 8 (80%) had died. Of 3 SCC patients, 0 were alive and 3 (100%) had died. The mean overall survival for all WHO subgroups was 52,2
months. The median survival for each subgroup was 48 months for TC, 59 months for AC, 30 months for LCNEC and 1 month for SCC. The median tumor size for all groups was 9 cm, with a median tumor size of 11 cm for TC, 8 cm for AC, 12 cm for LCNEC and 12 cm for SCC. Tumor size did not correlate to diagnosis in this cohort (Fig. 6).
Figure 6: Box plot of tumor size among WHO subgroups. Although the size average increases slightly from AC to LCNEC to SCC, the range of tumor sizes overlaps exceedingly. TC tumor size average is greater than that of AC and LCNEC.
4.2 Relationship of Proliferation Markers with Overall Survival 4.2.1 HE and Ki67 Slide Preparation
All formalin-fixed and paraffin-embedded tumor blocks were cut with a microtome and mounted on microscope slides. To generate HE slides, the HE-staining machine Tissue Tek Prisma, incubated the tumor-mounted slides according to the protocol listed in chapter 3. Ki67 is a common proliferation marker expressed in replicating cells during late G1, S, M and G2 phases of the mitotic cycle (Khan et al. 2013). Ki67 stains were conducted on tumor-mounted microscope slides by the Dako Omnis advanced staining system in the immunohistochemistry lab of the UMG. A detailed staining protocol is listed in chapter 3. The HE slides and ki67 slides were reviewed to
All formalin-fixed and paraffin-embedded tumor blocks were cut with a microtome and mounted on microscope slides. To generate HE slides, the HE-staining machine Tissue Tek Prisma, incubated the tumor-mounted slides according to the protocol listed in chapter 3. Ki67 is a common proliferation marker expressed in replicating cells during late G1, S, M and G2 phases of the mitotic cycle (Khan et al. 2013). Ki67 stains were conducted on tumor-mounted microscope slides by the Dako Omnis advanced staining system in the immunohistochemistry lab of the UMG. A detailed staining protocol is listed in chapter 3. The HE slides and ki67 slides were reviewed to